Lessons Learned from Fukushima Related to Seismic · PDF fileLessons Learned from Fukushima Related to Seismic Hazards at U.S. Nuclear Power Plants and Future Seismic Hazard Studies

Lessons Learned from Fukushima Related to Seismic Hazards at U.S. Nuclear Power Plants and Future Seismic Hazard Studies

October 7, 2014

John G. Anderson Professor of Geophysics

University of Nevada, Reno

10/7/14 1 US NRC: Lessons learned from Fukushima


--

.{}

ltU

~·~.attU .,.8 oaa

~b'?Jf~ll·')

:a~~-.•·uf:J! (ttUifT

x• • ....,.7

"'ut~u1-2,.., •• . ...... .!

Tohoku, 2011, M9

Results

10• Fau~ Distanc• (km]

10


Probabilistic Seismic Hazard

Maps

2 10- 1

~ a: Q)

~ 10-2

"0 Q) Q) u Jj 10-3

~ ::::J c

Hazard curve

Ac(r)

~ 10-4 1r'-"-"'-'-'=='-'-'-'-'-'='-'-'-'-'-'-~~~~-'-'-'-'-'-'==="'-'-'<· -San Francisco - "-c such that P = 2% forT= 50 years

1 o-5l.'::::;::::;============:c:::::_..__._j___1___J 10-2 10-1 10°

Y= Peak Acceleration (g)

Hazard curve "conceptually observable"


Probabilistic Seismic Hazard

Maps

.$ 10- 1

a:s a: Q)

~ 10-2

"0 Q) Q) (.)

~ 10-3

a:s :::::1 c

Hazard curve

Xc(Y) . .

~ 1 0 -4 r-r""""~"'---'-="""'---"'--'"'--'-'-=~---'-==.:.........--"-'-""""'. -San Francisco _ A.c such that P = 2% forT= 50 years

10-5~~============~==========~~~ 104 10~

Y= Peak Acceleration (g) 0

Hazard curve "conceptually observable"

Contour these points. Ac(Y)

Quality of hazard map depends on quality of the hazard curve.


50"

..a·

30"

·110" · IOCr

Seismicity model

Science Behind PSHA

.go·

Hazard curve

~ 10-1

C1l a: Q)

~ 10-2

"0 Q) Q) u Jj 10-3 (ij :l c:

~~.:: 3.000 1.000 0.300 0.100 0.030 0.010 0.001 0.000

·8()"

I ~ 01

001

..'i 1 0-4 t,.----------------- San Francisco - ' 'c such that P = 2% for T = 50 years

1 0-5'-'--~--:------=~----~-:-'---' 10-2 10-1

Peak Acceleration (g)

Ground motion prediction equations

Lesson from pre-2011 Hazard Map

• M9 earthquake off the north coast was not included in the model.

– Smaller fault segments expected

– Linkage not considered

• Lesson: need to include fault linkages.

– US National Seismic Model for 2014 does this in California – overcomes this problem.

– Needs to be extended to the rest of the west.


Lesson from lost credibility

• Main “evidence” criticizing PSHA – M9 in Tohoku, Haiti, Wenchuan, Christchurch

– Criticism misplaced

• Lesson: – To assure quality PSHA, need an open process

– Ongoing two-way communication with global seismology community

– Goal: pathway for new relevant discoveries be promptly considered, without even waiting for formal updates of hazard maps.


Process: U.S. National Seismic Model

• Collaborative, community model

– USGS internal & external research programs

– NRC, EPRI, NSF, PEER, SCEC, …

• Update every 6 years

– Inclusive regional workshops


National Seismic Model: Issues

1. Better models for uncertainties

2. Selection of ground motion prediction equations

3. Validation and utilization of synthetic seismograms

4. Basins and long period ground motions

5. Test hazard curves – especially at low probability

6. Time dependence

7. Induced seismicity


National Seismic Model: Issue 1

• Better models for uncertainties

• Why? – At low probabilities, uncertainties dominate hazard

estimates, and small changes in uncertainties have a strong effect on results.

• How? – Broadband seismic networks

– Deal with the diversity in earthquake sources

– Ground motion prediction (regionalized)

– Seismic stations at the nuclear facility for local effects.



• Validation and utilization of synthetic seismograms

• Why? – Where data is sparse, simulations, if credible, can help

reduce uncertainties.

– Many uses in Japan

• How? – Seismic networks provide data to determine Earth

structure and source domains and characteristics.

– Validation exercises



• Time dependence • Why?

– Some nuclear facilities are near active faults likely to rupture within the facility lifetime.

– Example: Palo Verdi hazard at long periods is dominated by southern San Andreas fault, highly likely to have a an M8 class earthquake in next 30 years

• How? – Paleoseismology research – Appropriately modify the hazard assessment



• Induced seismicity

• Why? – It is happening

– Could happen in more places

• How? – Depends on

human activity

– Difficult issue


Thank you


Extra Slides

10/7/14 US NRC: Lessons learned from Fukushima 15


• Selection of ground motion prediction equations

• Why?

– Current process too strongly influenced by judgment

• How?

– Analytical methods to explore GMPE space

– More explicit acceptance criteria



• Basins and long period ground motions

• Why? – Some geological structures can amplify ground

motions enormously at long periods. (e.g. Mexico City, Hokkaido, Las Vegas, Los Angeles)

– Engineers asking for this

• How? – Broadband instruments for validation data

– Modeling is believed most reliable at long periods



• Test hazard curves – especially at low probability

• Why? – At low probability, models are

most sensitive to uncertainty

• How – Old fragile geological structures

that have not been damaged by past earthquakes.

– Example: precarious rocks near Yucca Mountain with ages >10,000 Inconsistent with the PSHA.



(a)

Lessons from Fukushima: Diversity in Earthquakes

--=ii!B B

2011~3-09 2011~3-09(2) 2008~7- 19 2005- 12~2 2003-10-31 (a)

75 L--2....JOOL:------ll 5":-0---....LI 00-:-----5-LO ~ , •• Distance from trench j , •.

Sources: Lay et al. {2012), Ye et al. {2013}, Kanamori {2014)

Acronyms

• PSHA: Probabilistic Seismic Hazard Analysis • GMPE: Ground Motion Prediction Equation • M9: Magnitude = 9 • USGS: US Geological Survey • NRC: Nuclear Regulatory Commission • EPRI: Electric Power Research Institute • NSF: National Science Foundation • PEER: Pacific Earthquake Engineering Research

Center • SCEC: Southern California Earthquake Center

10/7/14 US NRC: Lessons learned from Fukushima 20

Documents

Lessons Learned from Fukushima Related to Seismic · PDF fileLessons Learned from Fukushima Related to Seismic Hazards at U.S. Nuclear Power Plants and Future Seismic Hazard Studies